.png)
Power Your Growth
hello@propensity.com
By Kelly Greenwalt
We left HubSpot’s UNBOUND 2026 conference with pages of notes, plenty of ideas, and one overarching takeaway: AI is changing go-to-market incredibly quickly, but the companies that win won’t simply be the ones using the most AI. They’ll be the ones that build the best systems around it.
That distinction matters.
Across UNBOUND, AI was everywhere. HubSpot framed the agenda around questions like how teams can actually use AI to drive revenue, what a modern GTM engine looks like, and how businesses need to adapt as buyer behavior changes. The agenda included nearly 300 sessions, with topics ranging from AI agents and signal-led selling to CRM decision engines, modern B2B growth and the changing customer journey.
But underneath all the new terminology (GTM AI, Revenue AI, agentic GTM, AI agents), we heard a much more fundamental conversation happening.
We’re moving from tools that help us execute tasks to systems that help us make better decisions.
And that has big implications for Marketing, Sales, RevOps and the technology connecting them.
One of our favorite ideas coming out of the week was something our colleague Davis Potter from ForgeX highlighted: GTM technology is becoming recursive.
Think about how most GTM technology has traditionally worked. We define an ICP. We build an audience. We score accounts. We launch campaigns. Sales follows up. We measure the results.
Then, hopefully, someone eventually looks at what happened and adjusts the strategy.
The emerging model is different.
A recursive GTM system continuously moves through a loop:
context → decision → execution → outcome → learning → better next decision
That idea was at the center of Propensity CEO Sumner Vanderhoof’s session, “Build an AI-Driven GTM Engine (Step-by-Step),” which was part of UNBOUND’s Deep Dive programming.
Sumner laid out a five-stage operating model:
Identify → Launch → Validate → Pursue → Measure
The important part isn’t simply automating those five stages. It’s connecting them so that the outcome of one cycle improves the next. As Sumner put it, disconnected data, campaigns, sales activity and AI tools don’t magically become an operating system just because we automate them. A GTM engine needs a repeatable path from market evidence to revenue learning.
That leads to a deceptively simple benchmark for whether your AI strategy is actually working:
Your first cohort should be your weakest cohort.
If the system is learning, every subsequent cohort should benefit from what came before it.
That’s a much higher bar than “we added an AI agent.”
This may have been our biggest takeaways of the entire conference.
There is tremendous pressure right now to find places to insert AI into GTM. And there should be. But Sumner made an important distinction in his sessions: build the process first, then let AI strengthen the decisions inside it.
AI can accelerate research, audience creation, messaging, prioritization and analysis. But if your ownership, evidence standards and handoffs are unclear, automation simply moves that uncertainty faster.
Our strategic work heading into UNBOUND kept coming back to the same idea. The objective isn't another standalone tool. It's a GTM process Marketing and Sales can both understand, trust and use—from deciding where to focus through measuring what actually happened.
This is also why we found the broader UNBOUND agenda so interesting. Sessions included topics like “The Decision Engine: How to Build a CRM That Drives Action,” “Agent Hub: Put Agents to Work Across Your Customer Journey,” “From AI Experiments to Automation That Converts,” and “The New GTM Engine for the Modern Business.”
The conversation has clearly moved beyond Can AI do this task?
Now we're asking: How should all of these AI-powered decisions and actions work together?
That’s a much more interesting question.
Another takeaway from Davis & Sumner that resonated with us: the days of treating manually weighted account and contact scores as sophisticated intelligence are numbered.
Traditional scoring asks us to assign values to behaviors:
Visited this page? +10.
Downloaded this asset? +15.
Fits our target industry? +20.
Hit the magic number? Send it to Sales.
But we now have access to dramatically richer context: CRM activity, website behavior, sales conversations, firmographics, contact roles, third-party intent, technographics, hiring activity, company changes, renewal timing and more.
AI gives us the ability to interpret those signals together instead of pretending each exists independently.
But Sumner's framing adds an important caveat: a signal creates a hypothesis, not proof.
Instead of one opaque score, we should be thinking about confidence across dimensions like ICP fit, buying motion, buying stage, audience activity and timing. And a good system should be able to tell us not only why an account deserves attention, but also what evidence is still missing.
That second part may be more important than the score itself.
This sounds obvious when you say it out loud, yet a surprising amount of B2B technology has historically treated the account as the primary unit of intelligence.
Sumner's second session, “From Signals to Pipeline: ABM as a Revenue System,” challenged that model directly.
A large company can contain multiple business units, multiple buying groups and multiple initiatives at completely different stages. Saying “this account is interested” doesn't tell a seller very much.
The better questions are:
Which people are involved? What initiative might they be connected to? What evidence supports that hypothesis? What should we do next?
This is why buying groups finally feel operational rather than theoretical.
Teams can build audiences top-down from target accounts, bottom-up from relevant people, around psychographic characteristics like professional skills and interests, or outward from known website visitors into the surrounding buying circle.
That creates a fundamentally different approach to personalization, too.
Instead of: Account → ad
We can consider: Buying role × channel × message × timing
The CFO, Marketing Ops leader and Demand Gen leader at the same company shouldn't necessarily see the same message—or even interact with us through the same channel.
The account creates context. The people create precision.
We spend a lot of time talking about Marketing and Sales alignment. And we’ve all heard this story before, right? Now it has additional context.
UNBOUND reinforced for us that the most practical place to fix it is the handoff.
One line from Sumner’s session captures the problem perfectly: Sales is not demand generation.
Marketing activity creates evidence. But someone, or increasingly, some combination of AI and human judgment, needs to determine whether that evidence actually indicates a buying initiative before asking a salesperson to invest meaningful time.
Otherwise, we're essentially asking Sales to determine whether Marketing's signal was real.
That damages more than efficiency. It damages trust.
Send enough false positives to Sales and sellers learn that the system isn't useful.
The alternative is validation.
Separate what is known from what is inferred and what remains unknown. Then route the account appropriately:
Pursue. Nurture. Disqualify. Research.
Not every signal needs to become an opportunity. And not every unknown needs to become a “no.”
Sometimes the best next step is simply to learn more.
That’s why we particularly liked Sumner's discussion of human market research. When digital signals are interesting but insufficient, talk to practitioners close enough to the problem to understand the current reality, friction, ownership and timing. The goal isn't to disguise prospecting as research; it's to gather enough evidence to route the account intelligently.
One of Davis's takeaways was that ABM is not simply pushing paid media to a target account list.
We couldn't agree more.
The more interesting model is a unified Account-Based GTM strategy spanning Marketing, Sales and Customer Success, with coordinated channels and an intentional Target Account Portfolio.
Sumner's sessions effectively described the operating system underneath that idea.
Identify who matters.
Launch coordinated experiences.
Validate whether a real initiative exists.
Give Sales the context and next action required to pursue it.
Measure what happened.
Then feed that learning back into the next cycle.
The second presentation distilled the loop even further:
Signal → evidence → route → seller action → outcome → learning.
That is much closer to a revenue system than a marketing campaign.
And it changes measurement. Impressions, clicks and engagement still matter, but they're evidence—not the destination. A connected system should eventually trace activity through validation, seller adoption, qualified opportunities, pipeline, revenue and seller effort.
Another point Davis made that deserves more attention is account economics. We spent some time at the Propensity booth discussing this idea, we’d love to hear your thoughts.
Marketers can build increasingly sophisticated campaigns. They can personalize to individual buying-group members. They can add research and validation. They can orchestrate Marketing and Sales touches.
But just because they can invest heavily in an account doesn't mean they should.
A company with a $50K ACV probably shouldn't blindly deploy an extremely resource-intensive 1:1 ABM motion across eight named accounts.
Your Target Account Portfolio should reflect the economics of the opportunity: expected value, resources required, available capacity and the appropriate deployment model.
AI makes personalization and orchestration cheaper and more scalable, but it doesn't eliminate the need to understand the math behind your GTM strategy.
Better targeting without economic discipline is still inefficient growth.
For all the talk about autonomous agents, one of our favorite messages from Sumner was surprisingly human: Sales should sell, not sequence.
Once an initiative has been validated, sellers shouldn't receive another research assignment. They should receive the account and initiative context, relevant stakeholders, evidence, channels, tasks and one clear next action. Ideally, that information should appear inside the CRM, sales-engagement or conversation tools they already use.
AI can maintain context. It can identify missing roles. It can recommend next actions. It can prepare tasks and capture responses.
But relationships still belong to people.
That same principle showed up in how we approached our UNBOUND GTM Skills. The model isn't “AI does everything.” Our five reusable workflows map directly to the GTM process—Identify, Launch, Validate, Pursue and Measure—helping teams turn strategy into repeatable AI-assisted decisions.
There is a progression here.
Start with AI-assisted decisions.
Then, as your evidence, integrations, rules and governance mature, move toward connected orchestration.
Sumner described that evolution as moving from simple AI, where humans provide context and review recommendations, to advanced AI that connects systems, monitors changing evidence, refreshes confidence and messaging, recommends routes and actions, and applies approved rules while escalating exceptions.
That feels much more practical to us than jumping directly to “fully autonomous GTM.”
There was also something especially meaningful about attending UNBOUND this year.
On the first morning of the conference, we announced that Propensity ranked No. 70 on the 2026 Inc. 5000, up from No. 2,591 in 2025. That puts Propensity among the top 100 fastest-growing private companies in America and the fastest growing Marketing technology company.
It's an exciting milestone. But it also gave us an interesting lens through which to experience the conference.
The market isn't asking for more disconnected technology.
Teams already have data.
They already have signals.
They already have CRM systems, engagement platforms, advertising channels, AI tools and dashboards.
The bigger opportunity is connecting those pieces into something that actually helps Marketing and Sales make better decisions together.
As Sumner said in our Inc. 5000 announcement, the future isn't about adding another disconnected tool or generating more signals. It's about creating an intelligent system that understands those signals, determines the appropriate next action and orchestrates action across the teams and channels responsible for revenue.
That statement could just as easily summarize my week at UNBOUND.
For years, marketing technology has largely helped us do more.
More campaigns. More audiences. More signals. More personalization. More content. More activity.
AI makes it possible to multiply all of that again.
But we’re increasingly convinced that more isn't the interesting opportunity. Better is.
Better evidence.
Better decisions.
Better coordination.
Better handoffs.
Better timing.
Better experiences for the actual humans inside a buying group.
And, importantly, a system that learns whether those decisions were right and gets better the next time around.
The GTM platforms that emerge over the next few years won't simply tell us what happened. They won't just score an account and hand us another dashboard. And they won't bolt an AI assistant onto an old workflow and call it transformation.
They'll increasingly connect signals, context, decisions, execution and outcomes into a continuous learning loop.
That was our biggest takeaway from UNBOUND 2026:
AI isn't the GTM strategy.
Building a smarter, connected, continuously improving revenue system is.
And we're only at the beginning.
Let’s go!
%201.svg){kind=icon}